There are a number of runtime tests that checks that a command fails as
expected, so add an assertRunNotOk() similar to the existing assertRunOk()
to handle that instead of open coding it everywhere.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Reviewed-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since NetworkManager 1.54, packaged since commit 24827ae398
(package/network-manager: bump to 1.56.0), new services related to
running in an initrd are now installed, which prevents proper startup,
with systemd reporting that two units declare the same BusName.
When systemd is configured without initrd support, drop the
NetworkManager initrd-related services.
Fixes: 24827ae398
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Fiona Klute <fiona.klute@gmx.de>
Cc: Marcus Hoffmann <buildroot@bubu1.eu>
Acked-by: TIAN Yuanhao <tianyuanhao3@163.com>
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Security fixes:
CVE-2026-53613 - mount(8) TOCTOU race on target path.
The SUID mount does not pin the mount target directory, allowing a
race between path resolution and the actual mount syscall. A local
attacker can swap an ancestor directory component between these
steps to redirect a mount to an arbitrary location.
Reported-by: Xinyao Hu
CVE-2026-53612 - mount(8) TOCTOU race on post-mount owner/mode change.
The X-mount.owner, X-mount.group, and X-mount.mode options use
path-based lchown()/chmod() after mounting. An attacker can swap
the target between mount and the ownership/mode change to gain
control of arbitrary files.
Reported-by: Xinyao Hu
CVE-2026-53614 - mount(8) SUID bypass via LIBMOUNT_FORCE_MOUNT2.
The environment variable LIBMOUNT_FORCE_MOUNT2 is not filtered
via safe_getenv() in SUID context. A local attacker can force
the legacy mount(2) code path, which uses a two-step bind+remount
or propagation sequence with a window where security flags (nosuid,
noexec, ...) are not yet applied.
Reported-by: Xinyao Hu
Full release notes: https://www.kernel.org/pub/linux/utils/util-linux/v2.41/v2.41.5-ReleaseNotes
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
With 8536a5d04a the upstream default
config is used, changing EnableNetworkConfiguration.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
In a standalone iwd setup having openresolv or systemd-resolved to
apply DNS config is useful, but if iwd is used with another network
management service (e.g. NetworkManager) the dependency is
unnecessary.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
Note this changes EnableNetworkConfiguration from true to false
according to the upstream default, users who want a standalone iwd
setup which handles DHCP will need to adjust the config. Adjusting
NameResolvingService according to selected packages is unlikely to
have unexpected effects, so keep that.
This makes iwd work as a drop-in replacement for wpa_supplicant with
NetworkManager.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
The built-in default of iwd is to log to stderr, which is lost when
sent to the background.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
* Explicitly set shell type. Shellcheck doesn't know OpenRC, but the
script as such is POSIX shell.
* Override warnings not applicable in context.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
With the action function as the last command in the script its return
code automatically becomes that of the script, and without explicit
exit shellcheck does not complain about unused functions.
Also wait for the process to stop in "stop", and simplify restart.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
If sending the stop signal failed for whatever reason, waiting for an
existing PID file to disappear is likely to block indefinitely.
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Tested-by: Andreas Ziegler <br025@umbiko.net>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
The exit status of the last command automatically becomes the exit
status of the script. And if there is no explicit exit shellcheck
accepts unused functions (here: start/stop/... called via variable).
Signed-off-by: Fiona Klute <fiona.klute@gmx.de>
Tested-by: Andreas Ziegler <br025@umbiko.net>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
libudev-zero version 1.0.4 added support for usb.ids with commit
2bebebc9e0
Also select hwdata when libudev-zero implementation for udev is used to
more useful output from usbutils. This is strictly speaking an optional
dependency but would be confusing if it weren't working.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
[Marcus: add note about selecting hwdata]
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
https://github.com/illiliti/libudev-zero/releases/tag/1.0.4
Removed patch which is included in this release.
Added buildroot-specific PREFIX to _BUILD_CMDS to override default
'/usr/local' in upstream Makefile to be compatible with the install
location of usb.ids provided by the hwdata package.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
Buildroot commit aa5e3eec63 bumped the
package to version 7.0 which includes commit
33b950714a
that adds an include of linux/time_types.h.
This header file was added in linux kernel version 5.1:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=ca5e9aba753ed15d173c7a7b88e4d402b7ca8121
Fixes build errors:
In file included from ../bundled/linux/include/uapi/linux/cgroupstats.h:20:0,
from netlink_nlctrl.c:11:
../bundled/linux/include/uapi/linux/taskstats.h:21:30: fatal error:
linux/time_types.h: No such file or directory
detected by the Gitlab pipelines with the following defconfigs:
- br-arm-full-static
- bootlin-aarch64-glibc-old
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Upstream commit
a00a3cce3c
added the code of gli & glm in-tree and removed the usage as external
packages.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Fixes build error not caught by the autobuilders with this defconfig:
BR2_x86_64=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_FORCE_HOST_BUILD=y
BR2_PER_PACKAGE_DIRECTORIES=y
BR2_PACKAGE_KODI=y
BR2_PACKAGE_KODI_SCREENSAVER_RSXS=y
BR2_PACKAGE_MESA3D=y
BR2_PACKAGE_MESA3D_GALLIUM_DRIVER_SOFTPIPE=y
BR2_PACKAGE_MESA3D_OPENGL_GLX=y
BR2_PACKAGE_MESA3D_OPENGL_EGL=y
BR2_PACKAGE_XORG7=y
BR2_PACKAGE_PYTHON3=y
BR2_PACKAGE_PYTHON3_PY_ONLY=y
BR2_PACKAGE_LIBUDEV_ZERO=y
Raised minimum version for all addons as requested by upstream.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Fixes:
* CVE-2023-7216 [1]
A path traversal vulnerability was found in the CPIO utility. This
issue could allow a remote unauthenticated attacker to trick a user into
opening a specially crafted archive. During the extraction process, the
archiver could follow symlinks outside of the intended directory, which
could be utilized to run arbitrary commands on the target system.
* CVE-2026-28525 [2]
SWUpdate contains an integer underflow vulnerability in the multipart
upload parser in mongoose_multipart.c that allows unauthenticated
attackers to cause a denial of service by sending a crafted HTTP POST
request to /upload with a malformed multipart boundary and controlled
TCP stream timing. Attackers can trigger an integer underflow in the
mg_http_multipart_continue_wait_for_chunk() function when the buffer
length falls within a specific range, causing an out-of-bounds heap read
that writes data beyond the allocated receive buffer to a local IPC
socket.
For full release notes, see:
https://github.com/sbabic/swupdate/releases/tag/2026.05
Remove patch that has been merged upstream for this release.
[1] https://github.com/advisories/GHSA-v9vx-4mxw-76j2
[2] https://github.com/advisories/GHSA-hggv-rg65-qf4h
Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Marcus Hoffmann <buildroot@bubu1.eu>
Our docker image was finally updated to Debian 12 and now provide Qemu
7.2.15 [1]. We can safely revert this workaround for older Qemu version.
[1] e5bbe020ca
This reverts commit 37e23e0dd7.
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
